Several project members are main organizers of short-term visual object tracking challenges, VOT. The aim of the challenges is large-scale distributed comparison of mid-level motion estimation techniques and establishing a common performance evaluation methodology in short-term mid-level motion estimation and tracking. The VOT project committee is international and includes partners from Slovenia, Great Britain, Austria, Czech Republic, Australia, and Sweden. The members of the project committee are lead methodologists, analysts and scientific coordinators. In addition to the challenges, they also organized two workshops at main computer vision conferences: VOT2013 at ICCV2013, and VOT2014 at ECCV2014. Both challenges as well as the workshops have received a highly positive response from the community. Over 50 authors have submitted their results to the VOT2014 challenge; the list included authors both from universities as well as industry. The results have been published in a 50+ co-author paper. Within the workshops, the project leader regularly gives invited talks on the challenges methodology and results. In 2014/15, the main open source library in computer vision, OpenCV, issued a call for several challenges in computer vision. For the tracking challenge, they have chosen the dataset that was developed as part of VOT2014 challenge.
B.01 Organiser of a scientific meetingCOBISS.SI-ID: 1536160963
Two members of the project group, dr. Ude and project leader dr. Leonardis co-organized workshop on active visual learning and hierarchical visual representations in conjunction with the major robotics conference ICRA 2014, held in Hong Kong in May 2014. The focus of the workshop was development of representations and learning mechanisms that provide capabilities to acquire new visual models and procedures in an incremental (developmental), open-ended way, grounded in the experiences of the robot acting in its working environment.
B.01 Organiser of a scientific meetingCOBISS.SI-ID: 28417319
Results of this project have been applied to an industrial project that the project group members are involved in. The project involves development of an automatic multimedia device for video-conference calls. Our client is developing a motorized stand for mobile phones that allows a user to select and track a person they are conferencing with while that person is moving around the room. We have developed fast and robust object motion estimation and tracking algorithms that are capable of running on a mobile phone and steering the motorized phone stand. To achieve that, we used fast motion computations and mid-level motion estimators, developed within this project. Such applications represents a major challenge both from scientific and application perspective, and are expected to prove useful in a variety of mobile applications beyond videoconference calls, giving our client a competitive advantage over other available solutions.
F.17 Transfer of existing technologies, know-how, methods and procedures into practice
Project group member, Janez Perš, became a member of the editorial board of International journal of distributed sensor networks (IJDSN) journal in year 2014. The journal has a JCR impact factor of IF=0.923. The scope of journal are distributed sensor networks, which includes visual sensors and related computer vision methods. The project group member’s editorial duties lie in the field of visual sensors and computer vision.
C.06 Editorial board membership
The group has developed a low-cost embedded camera that can be used as a visual sensor network node in inexpensive wired networks. The two distinctive features of this camera are the possibility of cost-efficient production even in extremely small volumes, and possibility of rapid development of computer vision algorithms that go beyond basic image processing found on most low-cost, low-performance embedded cameras. For these reasons, the camera, equipped with the software for analysis of human motion, was chosen as an important component of an intelligent environment controller, which is already commercially available on the market.
F.17 Transfer of existing technologies, know-how, methods and procedures into practiceCOBISS.SI-ID: 9897556